Has OpenStack failed?
I just republished this blog post examining the dissolution of OpenStack into echo chambers. What would have caused this failure if it has in fact occurred? Other large and roughly similar consortia, like Eclipse, have not succumbed to this dismemberment death. Perhaps it’s a characteristic of the makeup of the project, both in terms of the code and in terms of the financial, political structure. To me, and I readily admit my knowledge is imperfect, it seems to have gained the anatomy of a make-believe pony, at least after the first year, and has been driven more as a set of fronts for corporate leverage than as a consortium working on a commons. What holds a set of communities together is, I think, that very commons, that which is shared by all stakeholders.
But, perhaps I’m wrong. Yet there have been more and more declamations that OpenStack is ailing, even as it gains more and more corporate mass.
4 comments so far
Comments are closed.
newspeak 
I wouldn’t say that OpenStack has failed, it’s done exactly what it was designed to do. It was designed to generate a lot of code for other people to package and sell as a product. It’s designed to be a “meritocracy” with no “benevolent dictator”, and because of that it has no clear goal and no focus. OpenStack is whatever the consumer wants it to be, and that’s a lot of the problem I have with it. Even something like Eclipse has a focus (being a development environment for the Java ecosystem), but OpenStack is Public Cloud, Private Cloud, Telco NFV, “Free VMware”, Big Data, Orchestration, Authentication, Configuration Management, Container Consumer, Container Provider, and more. It’s succeeded at being a market juggernaut. In my mind it has not succeeded in producing good software, and it doesn’t want to because that could hurt the hand that feeds it. I got involved with OpenStack to build public clouds, but even the handful of groups who have managed to do that have done so with as much pain as possible, and the majority of loud voices in the ecosystem don’t care. That’s disappointing, but it’s not failure.
Carl, Well put. I knew a lot of people who were excited when OS was launched (I was one of them, there at the room at Oscon) and who then either worked in the community or wanted to (like me). We have then seen the relentless march toward a model that sacrifices vague notions of community in favour of catering to specific stakeholders with products they want to sell (in part by) using their engagement with OpenStack. I dislike vague notions of community and prefer specific conceptions of peer networks and their commons; and I am as aware as the next of the problems caused by gravitating (linking? chaining?) a developer community to the product needs of the sponsoring corporations. I did, after all, work for both Sun and Oracle.
Thanks for posting this! Very interesting observations. Your post makes me think that we need a general theory of Open Source Projects. We need someone to write _The Mythical Man-Month_ for the 21st century. Describe the ways that open source projects are born, grow, succeed, founder, fail.
@mick: Hm….It’s an attractive idea and I have thought certainly of doing something similar, originally for corporate-sponsored open-source projects, like OpenOffice.org, and also, of course, OpenStack. (Simon Phipps, a former colleague of mine at Sun and now with Wipro, has a book on open source and the enterprise; I forget the title, but Amazon surely has not.) The interesting thing about open source, however, is that its roots go deep, deeper in fact than the model of corporation making software that Brooks describes so powerfully. I’ve argued that much of what makes open source production and organisation interesting can be found in the early Rochdale Society of Equitable Pioneers (a form of which still exists), formed in 1844 in England. It has provided the model in the West for cooperatives and its emphasis on open collaboration and a commons was in part a stand against the ravages of community pulsed by the industrial capitalism transforming not just England but the world. (See the Rochdale Principles.)
By 1844, the idea of holding in common knowledge and not just land was as much a recollection of what had been lost (or taken: the commons) as a particularly modern effort to assert the necessity of doing (to use a modern term) in community.
But I’m being kind of evasive. I do think it’s reasonable to sketch if not a Brooks’ description of the modalities of open source, at least a set of basic elements expected in open source projects. Versions of these abound. See Apache, for instance; but also all the manifestoes that have been written since, roughly, 1998. But there is a point here to attend to. Open Source does not prescribe a method of doing. It’s like anarchism, in that: it’s anti-method. But not anti organisational.